Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
AI is in such an early stage
0:02
that the next five , 10 years are really
0:04
going to tell us what that industry is going
0:06
to look like . The more AI literacy
0:08
that you can build up today , the better . I
0:10
think the best way to do that is by following
0:13
your passions , your hobbies . You said you were working
0:15
with mid-journey . If you're an artist
0:17
, if you like to draw , if you like creating art , then
0:19
I think mid-journey is such a great tool to
0:21
start playing around with your own art and also
0:24
seeing how it responds to art . I
0:26
think AI literacy in order to develop
0:28
that the best way to do it is to figure out what are your hobbies
0:30
, what are your passions . If you like writing
0:32
, play around with chat to BT or other large
0:35
language models . If you like composing
0:37
music , or if you like paying music , play around with
0:39
an AI tool in the music scene and understand
0:42
how to prompt it , how to use it to create something
0:44
new . Because I think as you learn
0:46
those skills , you'll be able to become
0:48
an expert , because anyone today
0:50
can become an expert in AI because it's just so
0:53
new , so early , and the more time
0:55
you invest in it today , the better you'll
0:57
be . And the best way to invest time is by doing
0:59
it through a hobby or something that you're passionate
1:01
about .
1:02
Hello and welcome to Developers'
1:04
Journey , the podcast bringing you the making
1:06
of stories of successful software
1:08
developers to help you on your
1:11
upcoming journey . I'm your host , tim
1:13
Bologna . On this episode , I
1:15
receive Denzil Aden . As
1:18
a solo technical founder , denzil
1:20
carved her niche with
1:22
an AI-focused degree from MIT , mba
1:25
from Harvard and a distinguished
1:27
career at Microsoft . Wow
1:29
. She is devoted to making your life
1:32
smarter with AI , enhancing routines
1:34
, automating mundane tasks oh , I
1:36
love that one and maximizing every minute
1:38
of your day . Don't
1:40
love that one that much . As
1:43
with Smartie , and only in
1:45
one AI productivity assistant
1:47
. She builds , and when she unplugs because
1:49
she does sometimes Denzil
1:52
plays the piano , offers fiction
1:54
and upheads and subs tea
1:56
. That's a program
1:58
, denzil . Welcome to Dev Journey .
2:00
Thank you so much for having me . I'm excited to be here .
2:03
Oh , I'm excited as well , and it's been so long . We
2:06
scheduled a couple of times I had to reschedule
2:08
, so I'm really thrilled this is happening right now . Let's
2:10
wait Me too , but before we come to your
2:12
story , I want to thank the terrific
2:15
listeners who support the show . Every
2:17
month you are keeping
2:19
the Dev Journey lights up . If you
2:21
would like to join this fine
2:23
crew and help me spend more time
2:25
on finding phenomenal guests
2:27
than editing audio tracks , please
2:30
go to our website , devjourneyinfo
2:32
and click on the support
2:35
me on Patreon button . Even the
2:37
smallest contributions are giant
2:39
steps toward a sustainable
2:41
Dev Journey journey . Thank
2:44
you , and now back to today's guest
2:46
. So , denzil , as you know , the
2:49
show exists to help the listener understand
2:51
what your story looked like and imagine
2:53
how to shape their own future . So
2:55
, as usual on the show , let's go
2:57
back to your beginnings . Where would you place the
3:00
start of your Dev Journey ?
3:02
That's a great question and I
3:04
love the start of my Dev Journey because it started
3:06
very young . I was eight years old
3:08
, I was in third
3:10
grade and we had just taken our
3:12
first computer class where we
3:14
learned about logo . And I don't know if you're familiar
3:17
with logo , but it's this children's
3:19
programming language with a little turtle and
3:21
you tell the turtle where to go , what to do
3:23
to build a house , and
3:25
I fell in love . That's when I first
3:27
started programming . It's actually logo is like the
3:29
first time I used recursion
3:32
. It was the first time I learned how to make
3:34
if this , then that statements and I
3:36
fell in love and I clicked instantly
3:39
for me . I just understood
3:41
how to talk to this turtle
3:43
and I could tell that
3:45
my friends didn't get it as well
3:47
as I did and it just felt
3:50
very magical and I'm lucky . I grew
3:52
up in a very tech
3:54
area of California . I was in the Silicon Valley
3:56
, my dad was an electrical
3:59
engineer and a software engineer , and so I had
4:01
some exposure to that early on , but I
4:03
never thought of myself as a software engineer . That
4:06
was something that I even wanted to do until
4:08
I started playing around with logo and
4:10
it just opened up all of these doors for me and
4:13
I started coding because of that very
4:15
young . I took that passion with me through
4:17
high school to eventually
4:19
went to college at MIT and I
4:21
chose to go there because I really wanted to study
4:23
computer science , although I was still very
4:25
convinced that I didn't want to be a software engineer . I
4:27
wanted to do something else with technology . So I tried
4:30
biotech . I tried out a bunch
4:32
of different careers but I kept coming back to
4:34
just pure building and
4:36
I ended up doing a master's at
4:38
MIT as well . My master's had a theoretical
4:40
focus in AI , but
4:42
I did a thesis project in human
4:44
computer interactions and I ended
4:46
up building pretty much a precursor to Slack
4:49
, but for the classroom . So
4:52
it was a collaborative , asynchronous platform
4:54
, but for education
4:56
classrooms , and that's what I wrote my thesis on . It
4:59
was something that I wish I had stuck with , but
5:01
I really didn't think I could be a founder . I
5:03
thought that being a founder wasn't for me and
5:06
I wasn't really sure who the profile was . But it wasn't something that
5:08
I wanted to pursue . So I
5:10
ended up going to big tech instead . I
5:12
started as a PM at PowerPoint
5:15
, liked it but missed coding . So
5:17
I started teaching computer science at a
5:19
community college and then eventually at
5:21
San Francisco State , just as a guest lecturer . And
5:24
then I still missed coding and
5:27
so then ended up switching to being a software
5:29
engineer at Yammer . But I've always
5:31
been in the enterprise productivity space , always
5:33
in the future of work tools and helping people
5:35
be more productive at work , specifically
5:37
at Microsoft . But even during
5:40
that time I wasn't sure exactly what I wanted to do
5:42
. I wasn't sure if that was
5:44
what I should focus on for the rest of my life . So I started trying
5:46
out a bunch of different careers , from
5:48
law , from finance to politics , decided
5:51
to go to business school and really both broadened
5:53
my skill set but
5:56
also give me some time to figure out what
5:58
is it that I want to do with this time and
6:01
with my passion and my curiosities ? And
6:03
that's really where I came up with the idea for Smurdy . I
6:06
almost reverse engineered into it . I was still trying out different
6:08
things , but I asked myself if I could only work on
6:10
one thing for the rest of my life . What would
6:12
it be ? And this was back in 2018 . And
6:15
AI was interesting
6:17
, but it wasn't like it is today . And
6:19
it's funny . I used the same pitch for my startup , smurdy
6:21
, today that I did in 2018 , but
6:24
the appetite for it has changed so drastically
6:26
. But back then I was like , if
6:28
I do believe this future is coming and
6:31
I want to be part of that future . And so I started really
6:33
building Smurdy for myself , reverse
6:36
engineered into what the product
6:38
is today and tried to raise money in business
6:40
school . It didn't work out and
6:43
then I kept pitching to the same people . I
6:45
kept iterating my pitch , kept iterating the product and
6:48
eventually raised that first check which
6:50
made it easier to raise more money , and now
6:52
I'm working on it full time .
6:54
And congratulations on that . Thanks , but
6:57
that's a whole bunch to unpack before
7:00
we get to Smurdy . That's okay for you ? Yes , definitely
7:02
. I'd
7:04
like to come back to that turtle , the way
7:06
you talked about it , saying
7:09
well , I understood how to talk to that turtle . I love this formulation
7:12
Because that's really something . When
7:14
you're trying to talk to a computer and
7:17
trying to tell it to do something and you're not speaking
7:19
the right language , it doesn't click . As you
7:21
said , it's so
7:24
difficult to really Understand
7:27
the way you should be talking to it and the way
7:29
you personally find this turtle or this computer
7:31
as a total and say I understood how to talk
7:33
to it . I love it . It's really saying , hey
7:35
, you found the right language . Did
7:38
you realize right away the
7:40
power that software could
7:42
have and so what the
7:44
power you had learned in talking to that turtle
7:46
could bring to your life ? Not
7:49
yet at that point .
7:51
I don't know if at that point I understood how
7:53
transformative of a technology any
7:55
sort of software language , programming language is
7:58
, but I did understand that computers
8:00
spoke differently than humans did
8:02
, and that's what programming languages allowed
8:04
us to do , that they allowed us to communicate what
8:06
we wanted To
8:08
technology , and I actually think the greatest
8:11
thing about a I today is that
8:13
it's made it even easier for humans to communicate
8:15
with technology , because now technology can communicate
8:17
the way we communicate . But back then we
8:19
needed to learn how to use these
8:21
intermediary bridge languages in
8:23
order to be able to allow
8:26
software to do what we wanted , and
8:28
I thought that was really fascinating . I was always really
8:30
into logic and puzzles , and it was
8:32
just a puzzle that needed to be solved , and
8:35
then , as I grew older , I really
8:37
understood OK , like these are
8:39
the limits of technology , this is what are
8:41
the limits of my own programming skills , and try
8:43
to find connections there .
8:47
How is has been your , your
8:50
relationship with
8:52
students ? You mentioned you were a teacher for a bit
8:54
with students for whom it
8:56
didn't click the way you did for you and
8:58
and bridging that gap and helping
9:01
them understand exactly how
9:03
that work . If it clicked for you from
9:06
the get go love at first sight I would almost
9:08
say how did you bridge that gap and build this relation with students
9:10
?
9:11
That's a great question , and teaching was
9:13
like such a great experience for me because it really showed
9:16
me that everyone just learns in a different way
9:18
. No skill is unknowable . It
9:20
is just about finding the way that your
9:22
brain works , your thinking patterns
9:25
work , and trying to find alignment . And so
9:27
, for me , I really gave me
9:29
a great perspective on what is wrong with education
9:31
today , and also
9:33
that no one can say
9:36
that they're not a coder because you have the skills
9:38
, you have the ability to think . It's just
9:40
about figuring out how to make your
9:42
brain talk , the way that computer brains talk .
9:46
So do you have some , some , some techniques
9:49
, when it's not clicking , with people
9:51
to find the way that they need
9:53
to hear it , so that clicks ?
9:56
I don't know if it's necessarily about like how you
9:59
hear it , but I will say it is about practice , because
10:01
the more you do different
10:03
exercises , you'll start to recognize patterns
10:05
between how you solve certain problems
10:08
in the computer science space and
10:11
Java , and you'll find
10:13
the most problems that you're seeing in the
10:15
classroom setting are just variations of
10:17
the same same conundrum
10:19
. And then you just need to start recognizing
10:22
OK , like this is how these are the parameters
10:24
that the function needs . This is how functions
10:26
are usually structured , and so I think it's practice
10:29
more than anything else , practicing to think
10:31
the way that computers think and learning what
10:33
they need .
10:35
OK , so really using it , using it , using
10:37
it , and at some point you will
10:39
see and it will click .
10:41
Exactly , and I think that's true today
10:43
too , with all of these tools that are out
10:45
there and all of this , all
10:48
of the hype around prompt engineering . It is about
10:50
practicing learning how to prompt
10:52
these models . It's learning about
10:54
how AI models think
10:56
and understanding how you can , like , adapt
10:59
your own thinking to match that .
11:02
Isn't there a difference though , in
11:05
such that a programming language is
11:08
is very , very
11:11
Cartesian and definite . Either you
11:13
have the right way to ask or you don't , and there's no
11:15
two ways on all of those
11:18
, no three ways to do the same thing . In a way
11:20
, crafting
11:23
your prompts with chad , gpt
11:25
or with the LMS we have nowadays
11:27
, you can achieve the same results with
11:30
very different ways , and so dabbling into
11:32
it quite often leads to some
11:34
results as well , and it's not a black and
11:36
white result . It's very shades of gray , except
11:39
if you add some parameters etc . But but
11:41
if you stick to the , to the prompting
11:44
, it's really shades of gray . Harder
11:47
in computer science to double your
11:50
way into the , how the computer is supposed
11:52
to , to be talked to .
11:54
I think that's a great point . I agree with you . Like with
11:57
a traditional programming language
11:59
, there is only one right way to
12:01
talk to software
12:03
, to whatever you're trying to get
12:05
done with prompt engineering . Of course , you can
12:08
ask something in multiple ways and you'll get different
12:10
results , and you should probably experiment so you can
12:12
see what parameters lead to what results
12:14
. But at the same time , I do believe
12:16
that there's like an optimal way to
12:19
do prompting , and we just haven't fully
12:21
discovered that yet . I think five years from
12:23
now there will be textbooks that will say like this
12:25
is exactly how you should be prompting this model
12:28
specifically , and I think that
12:30
precision will come . But
12:32
right now we're all in this exploratory
12:34
phase , trying
12:36
to understand what is the best language .
12:40
Oh , we are still a baby phase
12:42
of asking questions . I'm
12:45
exploring a lot of mid journey
12:47
in the past days and weeks
12:49
and understanding
12:51
some patterns and how to have to ask questions
12:53
. This is , this is fantastic
12:56
. This is an endeavor in itself and
12:58
indeed , when you find something that really works
13:00
and you can reproduce and is really consistency
13:02
, oh , now I really learned something
13:04
. That's
13:06
very gratifying .
13:08
I agree it's an adventure and it's really fun .
13:12
It is so . You
13:15
said obviously MIT , obviously
13:17
computer science , even though you don't want to become
13:19
a hardcore software engineer
13:22
, you said but but still
13:24
it had to be in SIFT3 engineer . Did
13:27
you have doubts in which direction
13:29
to take inside
13:31
the CS space ?
13:34
No , I was always interested in
13:36
AI , so I knew that I wanted to do my theoretical
13:38
masters in that , and then I
13:40
always knew I wanted to pursue
13:42
computer science . There was no other major
13:45
that appealed to me . I just
13:47
wasn't sure what I wanted to do with that major
13:49
, and so that was what was most confusing
13:52
for me . And then , in terms of choosing
13:54
which university go to , I actually wasn't
13:57
sure I wanted to go to MIT , and
13:59
then I visited during their
14:01
campus preview weekend , which happens in
14:03
April , and it was the most
14:06
fun weekend I had ever had in my
14:08
life up until that point , and
14:10
it convinced me that this was the right place
14:12
for me to go to . And so
14:14
it all really just fell into place
14:16
, and I just needed to experience it once to
14:18
know it was the right fit .
14:20
Okay , was that weekend representative
14:23
for the years you did after ?
14:26
Yes , actually , it really was .
14:27
Okay .
14:29
It was very so . I think MIT has this motto
14:31
, which is work hard , play hard , and even that
14:33
weekend was very much like
14:35
extremely intense
14:37
but very fun at the same time , and
14:40
I think it just it does a really good job
14:42
of showing you what your college experience
14:44
could be like . It's all student run
14:46
. Obviously , there's like administration overseeing
14:49
everything , but students really plan
14:51
all of their events and it's just 24
14:54
seven for three days . Okay
14:56
, and the rest of the curriculum
14:58
is 24 seven for four years , five
15:00
years , but it's a little
15:02
bit less recreational
15:05
only , but I mean even , but it's the same
15:07
level of intensity .
15:08
Okay , that's good , that's good . It's really
15:10
a hard or a very thin line to
15:13
to meet during
15:15
those open doors to really show
15:17
the the values and what
15:19
the school is made or the college is made
15:21
of , and then have
15:23
the same feeling afterwards and not feel
15:26
lied
15:28
to . It's really hard , it's
15:30
a very hard balance .
15:31
I personally I'm obviously biased , but
15:33
I thought MIT did a really great job and I actually
15:35
went with a friend of mine from high school
15:38
. We both went and she did not have
15:40
as good of an experience and she did not end up going
15:42
to MIT and so I think it
15:44
does do a good job for prospective
15:46
students to come see what the atmosphere
15:49
is like , see what the culture is like , and then have them decide
15:51
like is this for you or not ?
15:53
Yeah , indeed , indeed , indeed . Then
15:55
, moving on , when you spoke about your theoretical
15:57
stereotypical thesis , you
16:00
just said I wish I'd stayed with
16:02
this topic . What
16:05
do you mean by that ?
16:06
Yeah , so it's actually not with . So
16:08
the way that my master's was broken
16:10
down I could take classes
16:12
and concentrated in a space and then
16:14
I could do a thesis project in
16:18
any space that I wanted , ideally
16:20
with complementary skills
16:22
to my theoretical thesis . Well , my theoretical concentration
16:24
wasn't AI , but back then it was . Machine
16:26
learning was there , but it was still early
16:29
. I didn't really like pursue that as much . I
16:31
was still doing more of like a high level
16:33
. AI 101 theoretical concentration
16:35
and then my application
16:37
thesis project was in human
16:40
computer interactions and I built this
16:42
platform called Nora no one revises
16:44
alone and it was literally
16:47
slack , but for classrooms and
16:49
it was for helping students asynchronously
16:52
collaborate on assignments during
16:55
the school day , and I , in
16:58
hindsight and even then , could have turned
17:00
it into an actual product and business
17:02
, but I didn't do it and
17:05
I don't know why . I
17:07
just felt very I guess I was very afraid of the idea of
17:10
striking out on my own and
17:12
building a company with a product that I
17:14
already had , with users who were
17:16
willing to pay for it , and I
17:19
think it was just . It was a different time really , because I feel like
17:22
universities today have a lot
17:24
more programs
17:26
around entrepreneurship and MIT than
17:28
did , but it wasn't a
17:31
big part of the campus culture and if
17:33
I go to campus now , it's so different . There are so many
17:36
programs around . How do you start a startup , how do you raise
17:38
money , how do you start building
17:40
something people want , and
17:43
so that's really what I meant . I had this thing that I really
17:45
, even in hindsight now , could
17:47
have created a real company around , but it was
17:49
just so hard for me to grasp that
17:52
that was something I could do and that this was enough
17:54
of a starting point to get started .
17:57
Okay , makes more sense . So you were not ready
17:59
for the startup growing experience yet
18:01
, yeah , okay
18:04
, and does this product
18:07
still has its place in the
18:09
world 10 years later ? No , timing
18:11
is everything .
18:12
And even then it was a
18:14
little bit too early for that
18:16
asynchronous collaboration
18:19
platform , but I definitely think
18:21
there was still a lot of value that it was adding , especially
18:25
in the classroom setting . But today there's so many alternatives there's Slack , there's
18:27
Miro , there's so many different platforms that you can use to
18:29
get that same experience . Honestly
18:33
, figma could even replace what I built , and
18:35
so I think it had its time . But that's time that
18:39
time has passed Okay , so
18:41
no regrets .
18:42
No regrets so then you brushed over your your
18:44
first years
18:48
PM at PowerPoint
18:50
, then think , yemmer . Then there
18:52
was one in between . I think how long did that
18:55
phase last ? Or
18:58
go until you had this
19:00
reboot of saying what do I want to do with my life ?
19:03
Yeah , I was working for three
19:05
years a little bit more than that and then I decided to go to business
19:09
school and business school was really that period of reflection .
19:13
So how did you three years go ? I
19:16
mean starting the three years entering the , the workforce
19:18
with big
19:20
air quotes or industry or big
19:22
tech , as you mentioned it . How did that go ? How did it relate to
19:26
the ideal you had
19:28
in your mind ? How did that work ? And then , how do these three years
19:30
evolve into you questioning this and say , well
19:32
, maybe I should go back to business school
19:35
or whatever . I'm very interested
19:37
in those two . To pals yeah , no , it's a great question
19:39
because I don't think that enough people talk about , like what
19:41
that
19:45
feels like when you first go into the workforce .
19:47
It was jarring for me . It was not what I
19:51
expected . I wasn't sure if I
19:53
loved my job and I didn't know why . And it turns out that
19:55
I really did like my job . I was just hungry
19:57
for more and I just felt I wasn't
19:59
being challenged enough . I didn't
20:02
love my commute , which is so funny now to talk
20:04
about , but my commute was terrible back then
20:07
. It was an hour because I lived in San Francisco
20:09
. I wanted the lifestyle of being young and happy with my friends all the time
20:11
, but then I was commuting to Mountain View
20:13
and with
20:18
traffic that would take an hour each way , I just
20:22
felt surprised at what
20:24
I did like and also surprised at what I didn't
20:26
like . I didn't realize how important the lifestyle
20:28
aspect would be to me . I didn't realize how
20:30
much I would miss coding and actually building something from
20:36
like specifications versus writing the specifications
20:38
, and I didn't understand how different
20:42
those skill sets were . I didn't really understand at that
20:44
time how in front of was to understand users and
20:46
be customer obsessed , even though I was working in product
20:48
and it was . I
20:50
was learning so much but at the same time not appreciating
20:53
what I was learning , and so it's funny
20:57
to look at it with hindsight . And then , because I missed coding so much , I started
20:59
teaching
21:02
computer science and I realized
21:05
that I liked teaching , but it wasn't something that I wanted
21:07
to do as my full-time livelihood , so I didn't
21:09
want to go back to school to become a professor
21:14
and so I started looking at other jobs
21:16
and I think my number one priority at that time had
21:21
been finding something that
21:23
was in San Francisco . So that was my number one priority . And
21:25
then I
21:28
was lucky enough to interview at Yammer , which was a acquired
21:30
startup at Microsoft and their headquarters was in
21:32
San Francisco . So that worked out really well
21:35
on that specification and they were looking for software
21:37
engineers . So it worked out in that way
21:39
and I still got to stand at the Microsoft
21:43
umbrella , so it was an easy transition and
21:45
so it worked out so well for everything that I thought I wanted at that
21:47
time . So I switched to being a software engineer at Yammer
21:49
and I actually really
21:53
liked it . I liked being the one building
21:55
and seeing my changes get deployed
21:57
into the product
22:00
and seeing how users use them . It was just such
22:02
a satisfying feeling . So I was like , okay , I
22:04
do like this part of it . But
22:06
then , a year into that , I
22:10
ran into that same feeling of like okay
22:12
, but I want more . I want more challenges , I want
22:14
more to do . I don't know if this is
22:16
where I want to be for the next five or 10
22:18
years . I
22:20
just felt really hungry
22:23
but also really lost . That's
22:25
why I decided that business school was the right choice for
22:27
me . But it's so funny I remember
22:29
even now , when I decided to go to business school
22:31
, the head of engineering at Yammer , who I like never
22:34
worked with but I knew of , sat
22:36
me down and said okay , I
22:38
see how hungry you are , I see that you're
22:40
like pursuing other things and I don't
22:43
think you need to stay at Yammer . This is not a pitch
22:45
for you to stay at Yammer . I can help
22:47
you work at any startup that you want
22:49
, any place where you can like build the skill
22:51
set , but I feel like you should really go and
22:53
build instead . That was the first time anyone had
22:55
said to me like going to business school was the wrong
22:57
choice , and I
23:00
really look back on it today very fondly
23:03
because I understand what he was trying to say . He was saying
23:05
there are so many ways to get the same
23:07
skill set and to feed
23:10
that passion of wanting to build and learn
23:12
at an accelerated rate . I
23:16
think it really stuck with me about wanting
23:18
to go to a different startup , like starting a startup going
23:20
somewhere younger and smaller . I
23:23
think I just thought about that a lot while I was at business
23:25
school . That was my experience working
23:28
full-time for the first time .
23:31
It makes a lot of sense . How did
23:33
you come up to the idea of going back
23:36
with their quotes again to business school and
23:38
decide on business school that business
23:40
school was the right thing for you at that time
23:42
?
23:43
That's a great question . I actually applied to a lot
23:46
of different grad schools . I applied to law school , I
23:48
applied to a school for masters
23:50
of public policy . I
23:53
was confused in a
23:55
lot of ways about what I wanted
23:57
. The great thing about tech is that
24:00
it can be applied to any industry . There are
24:02
law tech , there is finance tech
24:04
. There is so many different ways you
24:06
can apply technology to different industries . I
24:08
was like , okay , I can bring this skill set to
24:10
any space and any career that I want
24:12
. I just have to figure out what I want to do . I
24:15
kind of glossed over this , but in my working
24:18
days I was actually interning on
24:21
the side at a lot of other things as well . I
24:23
was working for a Kamala
24:26
Harris Senate campaign . I was volunteering
24:28
for them and helping them with fundraising
24:30
. I was interning with a
24:32
criminal lawyer and trying to understand
24:35
if I wanted to go into criminal law , and this was
24:37
actually my third time trying out
24:39
a law internship . I had already worked in patent
24:41
law . Before that I tried a corporate law
24:43
internship as well . I
24:46
was confused and I was curious
24:49
. I wanted to try a lot of different things
24:51
and figure out what was right
24:53
for me .
24:55
I love the experimental approach
24:57
Really . Okay , I don't know
24:59
what I want , so let's try this , let's
25:02
try that , let's try that , let's try that and see what sticks
25:04
. Very entrepreneurial
25:06
.
25:07
Definitely . It's so funny because I never looked at
25:09
that skill or that the
25:12
mindset as entrepreneurial . But now that
25:14
I look at my life
25:16
in hindsight I can say very clearly
25:18
well , those are entrepreneurial skills . That
25:22
is part of it , that experimental
25:24
learning by doing which is not for everyone . Some
25:26
people are actually really great at looking at other people's
25:29
experiences and other people's mistakes and
25:31
learning from that . I'm not , and I wish
25:33
I was better at that , but I am very much a
25:35
experimenter , trier , iterator
25:38
.
25:40
Which you need nowadays , so that's fine . So
25:43
I've reached the point
25:45
in time where Smarties started to evolve in your
25:47
mind .
25:48
Exactly .
25:49
So can you tell us
25:52
the birth story of Smarties
25:55
before it became the company ? How did
25:57
you come up with the idea ? How did you did
25:59
that ? Maybe scratch an itch you had
26:01
or really started to evolve
26:03
into , from maybe just an
26:05
idea or a hobby into hey
26:08
, this could be more . And
26:10
what the first steps look like .
26:11
Yes , definitely , it's
26:14
very clear in my mind . So
26:16
it happened during my second year of business school
26:18
, maybe a little bit during the first year
26:20
I'd all . I arrived
26:22
at business school , I liked it , I
26:25
was open to trying out other careers
26:28
, and then I kept finding myself coming back to tech
26:30
and I said that I didn't want to pivot out of technology
26:32
and so what can I do with
26:34
technology ? And
26:37
HBS was really the first time I felt supported
26:39
in entrepreneurship and
26:41
pursuing that as a journey . There
26:44
were so many classes around it . I still think HBS
26:46
is very theoretical when it comes to entrepreneurship
26:48
, compared to some other business schools that are , I think , a little
26:50
bit more application heavy , and I
26:53
know they're changing that . They're trying to
26:55
become more application and
26:57
experimental focus , but at the time it was
26:59
still more theoretical , I think . But
27:02
it was my first exposure to that as
27:04
an experience , and one of my closest friends
27:06
at a business school had been a founder before
27:09
, and so I really got to see what that journey had
27:11
been like for him and he wanted to start another
27:13
startup right away , and so
27:15
it was exciting to see the
27:18
world through his eyes and that lens
27:20
and it definitely encouraged me to
27:22
think about that as a career path
27:24
for myself . And so then I started asking myself
27:26
okay , if I could only work on one thing for the rest
27:28
of my life , what space do I want to
27:30
be in ? And AI had always been something
27:32
that I was really excited about . I wanted to help be
27:34
part of that journey of making that a reality
27:37
, and back then it was still very much machine
27:39
learning , natural language processing , nothing
27:41
like what we have today with LLMs
27:44
, but it was still something that was rapidly
27:47
advancing and so I was like I want
27:49
to be in that space , I want to see how I can get
27:51
into that . And then I started asking about questions
27:54
in my own life , problems in my own life , and
27:56
at that time I really vividly remember I
27:59
was having anxiety
28:01
attacks like pretty much every day . I just felt
28:03
so overwhelmed with everything that was in
28:05
my life , both my personal commitments , my
28:07
professional commitments , everything
28:10
that I was trying to balance , and
28:12
I had this realization that
28:14
more than 25%
28:16
of the things that I was doing every day could easily
28:19
be automated with existing technology
28:21
. And I remember looking
28:23
it up and I think a lot of corporate
28:26
knowledge workers at that time had like 30%
28:28
. Today I would say it's about 60%
28:30
of the work that we do every day could easily be automated
28:33
with existing technology , and the big problem
28:35
there and why that wasn't being solved
28:37
with technology , was that none of
28:43
our software was able to talk to each other . We
28:45
had to manually handle every software
28:47
individually . Everything was in these different silos
28:50
, and so I was
28:52
on this kick of OK
28:54
, everything has an API . Apis can talk
28:57
to each other . Apis are programming
28:59
languages . Apis are languages for
29:01
how software speaks to other software , and
29:03
all I needed to do was create this
29:05
hub that allowed these APIs to connect
29:08
, and I should be
29:10
able to talk to this hub , and so that's how Smarty
29:12
really came up as an idea in my
29:14
mind . I had this vision around this
29:16
conversational web operating system
29:18
where I tell Smarty what I need to do , smarty
29:21
figures out the intent , what APIs
29:23
are relevant , and then connects it all together and makes it
29:25
work , and that's
29:28
how I came up with the idea , and
29:30
I started building Smarty as a chat
29:32
bot , just something that I was chatting
29:34
to . I would tell Smarty things that were on my
29:36
plate , and if Smarty could automate it
29:38
, it would , and that's how I got
29:40
started . That's the pitch that I gave
29:43
to my first investor . I
29:45
didn't really understand anything about how
29:47
to pitch , how
29:49
this is too big of an idea and you need a niche
29:51
to get started . You need a real
29:53
pain point , you need a real customer . It was definitely
29:56
a learning journey for me , but
29:58
that's how I got started .
30:00
And it is awesome , it was very long-winded . No
30:04
, no , that's . It's really cool . I was really thinking
30:07
what would be the analogy with existing
30:09
tools . It's kind of an
30:12
overlay over Zapier , which
30:14
has the connectors , but not this AI
30:17
part , where it figures out on
30:19
its own what it should be doing . You
30:21
really have to tell them , but
30:25
it's not just that . So , yeah , interesting
30:27
. And this started in 2018
30:30
, isn't it so
30:32
? Before chatGPT emerged
30:35
, or at least to the public , before
30:38
everybody understood
30:40
that this is happening ? How
30:43
was it to work in LLM back
30:45
then versus now ?
30:47
Yeah , it was a different way of thinking about
30:50
AI . It was very much around I was using
30:52
an open AI API
30:54
for natural language processing , but that isn't
30:57
what chatGPT is today . Then
30:59
I had these goals around creating machine
31:01
learning models , but everything that has happened
31:03
in the last couple of years has completely changed
31:05
that Not just for me , but for people who
31:07
were machine learning experts working at companies
31:10
building these models , it is
31:12
a brave new world , but not
31:14
a surprising one . Even back then , I had
31:16
that same vision that LLMs make even
31:18
easier today , but it's
31:21
just a different methodology to get there .
31:24
Did you change some part of your stack after
31:27
LLMs came out ?
31:34
No , I would call it AI 1.0 . We
31:36
had that before
31:38
and today we're still using that , but now we're layering
31:40
in AI 2.0 with all of the
31:42
language learning models and also large action
31:45
models . Really
31:47
, this is a new
31:49
time for how
31:52
the AI works , but in terms of
31:54
the actual infrastructure that you need to build
31:56
a product like Smarty , all of that stays
31:58
the same the APIs that
32:00
were connected to you , how those APIs connect together
32:03
, how you as
32:05
a user communicate with Smarty and what information
32:08
you need to specify all of that stays the
32:10
same . In some ways , I feel like it
32:12
was great that I had that period of building up that
32:14
infrastructure and now I get to
32:16
just play around with how the AI
32:19
at the top level works for how users
32:21
interface with Smarty .
32:24
Who do you define as your competitors ? Are
32:28
those the Zappiers which
32:30
are trying to add AI on top of
32:32
the services now , or is that a different space
32:34
?
32:35
It's a different space , and so I can
32:37
almost see Zappier or IFTHT IFTHT
32:40
, this and that being APIs that we integrate
32:42
with . It's really not even a
32:44
competitor because they
32:46
are really providing
32:48
the tools from creating these very intricate connections
32:51
, and I can see Smarty even using those
32:53
intricate connections between these different
32:55
APIs . Our competitors are more in the
32:57
future of work tool space , so tools
32:59
that are creating countering solutions , tools
33:01
that are creating email management solutions , contact
33:03
management solutions , and these are tools that
33:05
you're using at work . But
33:08
a lot of our users have product fatigue because
33:10
they don't want to switch between all of these different
33:12
niche productivity tools . They instead
33:15
want an all-in-one platform
33:17
that handles all of their favorite productivity
33:19
features , and so we're more in that
33:22
space , and so I would say Smarty's biggest
33:24
value proposition is first , this all-in-one
33:27
platform , trying to integrate with everything
33:29
across your administrative stack and right now
33:31
we're just on top of G Suite but everything from your
33:33
email to your Google contacts , to your Google
33:36
calendars , across multiple accounts
33:38
. So your professional accounts , your personal accounts
33:40
, just having all of that accessible in one place
33:42
. So that's our top value proposition . Our
33:44
second one is really around these conversational
33:46
commands , so being able to say something
33:49
like coffee with Tim at Blue
33:51
Bottle in San Francisco at 2 pm
33:53
London time and having that calendar
33:55
event sent with the right time zone with the right location
33:57
, just with keyboard shortcuts and
34:00
conversational commands . So that's our second biggest
34:02
value proposition . And the third one is using
34:04
AI to have recommendations
34:06
around when is the best time to get something
34:08
done , what is the best way to plan your schedule
34:10
? Brain-dumping tasks into Smarty
34:12
and having Smarty AI autoschedule it
34:14
into your day or into your weeks
34:16
and so really helping you make sure
34:18
you never drop the ball . And I think that last
34:20
value proposition is really where most of our
34:22
competitors are . So there are a lot of
34:24
tools that are helping you use AI
34:27
to autoschedule your day , and that's
34:29
where our biggest competitors
34:31
are .
34:32
Okay , makes sense . How
34:34
did you tackle the
34:36
problem of trusting the AI ? I
34:38
mean , if I'm just conversing with
34:41
AI and I trust this AI , he's
34:43
going to create the right appointments , send
34:46
those to my counterparts . It's
34:48
going to be a different time zone and it's
34:50
going to do it on its own and
34:52
correctly . I would double-check
34:54
everything it's doing .
34:55
Yeah , no , of course , and that's actually one
34:58
of the biggest things that we tried
35:00
to keep in mind while building out Smarty
35:03
. So I don't know if you remember XAI
35:05
it came out before
35:08
2018 . It was this assistant
35:10
, amy . You would CC
35:12
Amy on your emails and then Amy would
35:14
just do things for you , and the problem
35:17
with that experience was exactly
35:19
what you said that you were like what
35:22
is Amy going to do ? How is Amy
35:24
going to do this ? What if Amy does
35:26
it the wrong way ? And so , really , our goal with
35:28
Smarty is not to create this black box
35:30
around how things are getting done , but instead
35:32
to give you the right
35:34
commands at the right time so you know exactly
35:37
what is being done and how it's being
35:39
done . And so , instead
35:41
of you just delegating tasks
35:43
away to Smarty , it's really taking
35:45
these workflows like switching to my calendar
35:48
app , dragging and dropping , adding this contact
35:50
to that calendar invite . Instead of doing that
35:52
at that crucial step where you
35:54
have all of these mundane workflows that you
35:56
have to do , smarty is specifically replacing
35:58
that experience with a single command
36:00
where you control everything . You control exactly
36:03
the time , you control exactly the people . So
36:05
really finding that balance between giving you the
36:07
right control versus delegating
36:10
the mundane parts of product
36:12
switching , figuring out
36:14
like how does this work , things like that ?
36:17
Okay , a hard
36:19
balance to find , but
36:22
when you find it , I really trust
36:24
it to
36:26
really free up your mind , free up your calendar
36:28
, doing stuff for you , but I'm
36:31
still yeah , I can see
36:33
that you're like I don't believe that you
36:35
can fully take away
36:37
that fear , and I think it's .
36:38
You have to try the product . But it's really
36:40
. You are in control , you trigger
36:42
the commands and you
36:45
know exactly how the command is going to
36:47
work . So that confidence , I think
36:49
, and that trust , builds up over time .
36:51
Okay , I'll have to give it a go . Then you
36:56
mentioned a couple of times you
36:58
were searching for something to work
37:00
on for the rest of your life . How
37:03
does that collides maybe
37:05
doesn't collide with the
37:07
stellar growth of AI
37:10
right now ? Will
37:12
there still be a need
37:14
for how
37:17
to put it a
37:19
tool that is deep enough in
37:21
terms of you telling it what
37:24
it should be doing and really helping it
37:26
, versus , at some point , some
37:29
AI able to discover that
37:31
on their own and basically over
37:34
patting your
37:36
tool on the right and just doing
37:38
it without somebody teaching them that ? Do
37:40
you see what I mean ?
37:41
Yes , I do , but I still think
37:43
in that scenario , there will always
37:46
be an interface that you're interacting with as
37:48
the user , and , regardless
37:50
of what the AI is doing
37:52
, there's something that you are interacting with as the user
37:54
, and I want to be part of building that platform
37:57
, that interface , whatever it is . This doesn't
37:59
have to be smarty , and so I
38:01
think one of the big things as a founder is
38:04
you'll have a lot of ideas
38:06
, but finding something that you can
38:08
really , that you're really committed to in
38:11
the long run , that's really important , and for me
38:13
, even if smarty fails this concept
38:16
, this vision of what that
38:18
interface looks like , someone is going to create
38:20
it , and if I'm not going to be doing it at smarty , then
38:22
I want to go find a competitor or
38:24
another company that's working on that same vision
38:26
and go work for them , and so that's really how
38:28
it aligns for me .
38:34
And that is awesome . It really makes a lot of sense
38:36
when you describe , or when we talked about , the
38:38
product piece and the trust piece
38:40
on top of the AI . So really
38:42
, this is something that I don't see
38:44
AI solving for us . What's
38:47
happening in the background ? Yes , maybe , but
38:49
this whole how humans or tailoring
38:52
for humans , is really a hard
38:54
nut to crack , and this I don't expect
38:56
the UI to crack that in
38:58
the near future . So that's where your
39:00
tool or maybe another , as you're saying really
39:04
has to shine and
39:06
come into play .
39:08
I definitely think it's . We're really early
39:10
, we're not sure what that's gonna look like
39:12
, but it's exciting to be part of that journey
39:14
that shapes it .
39:16
Very cool . Coming
39:18
back to your students , was
39:22
there something , some kind of piece of
39:24
advice , stuff you told them again
39:26
and again to help them kickstart
39:28
their career , start on the right track
39:30
and maybe
39:33
get a glimpse on what you were doing or
39:35
something else ? Is there such such an advice
39:37
?
39:38
Yeah , it's a great question . It's actually something I've been thinking
39:40
about a lot recently , especially
39:43
with the rapid
39:46
buildup of AI tools that are coming out there
39:48
. I think AI literacy is something that is becoming
39:50
more and more important than ever before
39:52
, and that doesn't mean understanding
39:54
how AI works behind
39:57
the scenes . It's about understanding how
39:59
to use AI as a tool to
40:01
improve your own job , whatever you are doing
40:03
, and it's something that I am
40:06
telling folks today who are interested . Ai
40:08
is in such an early stage
40:11
that the next five , ten years are really
40:13
going to tell us what that industry is going
40:15
to look like , and so the more AI literacy
40:17
that you can build up today , the better , and I think
40:19
the best way to do that is by following
40:21
your passions , your hobbies , and so you said you were working
40:23
with mid-journey . If you're an artist
40:26
, if you like to draw , if you like creating art , then
40:28
I think mid-journey is like such a great tool to
40:30
start playing around with your own art and also
40:32
seeing how it responds to art , and I
40:34
think , ai literacy . In order to develop
40:36
that . The best way to do it is to figure out what are your hobbies
40:39
, what are your passions . If you like writing
40:41
, play around with chat , tbt or other
40:43
writing language , large language models , if you like
40:45
composing music or if you like paying music
40:47
, play around with an AI tool in the music scene
40:49
and understand how to prompt it , how to use
40:51
it to create something new . Because I think
40:53
as you learn those skills , you'll
40:56
be able to become an expert . Because
40:58
anyone today can become an expert in AI
41:00
because it's just so new , so early , and
41:03
the more time you invest in it today , the
41:05
better you'll be . And the best way to invest time is to
41:07
by doing it through a hobby or something that you're
41:09
passionate about . I think that's the same advice
41:11
I would have given my students 10 years ago
41:13
in these classes . If you're interested
41:15
in technology , technology can be applied to
41:18
any space , to any industry . Even back
41:20
then , if you were interested in music
41:22
, you could try to build a piano using
41:24
software . You could try to build I
41:27
don't know like a drawing tool or an art
41:29
tool using software . I think
41:31
there is something really powerful
41:35
about being able to align your passions
41:37
with your interests
41:39
and your goals and your learning
41:42
skills , because the more you can tie in something you're
41:44
trying to learn to something that you enjoy doing , the
41:46
faster and more fun I'll be .
41:48
Amen to that and I love that your answer
41:51
digs into exploration
41:53
again , saying hey , do
41:55
something with it and try what
41:57
happens . And this is exactly the same discourse
42:00
you had at the beginning of the show , talking
42:03
about how you went at your life . So
42:05
it's all full circles .
42:07
I think life , you
42:10
know yellow , you only live once . So
42:12
you need to explore everything
42:14
and figure out what is what brings
42:17
true to you .
42:18
Then go explore Fantastic
42:23
. Where would be the best place to
42:25
continue the discussion with you ?
42:27
Please find me on LinkedIn . I'm Denzil
42:29
Eden and I love
42:31
talking to people about their
42:33
careers , their journeys , their interest
42:36
in technology , ai , literacy , and I'd
42:38
love to give anyone a personalized onboarding onto
42:40
Smarty , and then you can always try out Smarty at
42:42
wwwsmartyai
42:44
. We're an early product , we have
42:47
some customers and we're always looking
42:49
for more people to give us feedback .
42:51
Then you heard her let's go there , anything
42:55
else .
42:56
No , that's it . That's me in a nutshell .
42:58
Fantastic , denzil , it was really
43:00
fun . Thank you so much for sharing your life and for
43:02
the good love we had . I had
43:04
a fun four times I had fun too .
43:05
thank you so much for having me .
43:07
Like was , and this has been another episode
43:10
of Devil's Journey . It was each other next
43:12
week . Bye , bye . Thanks
43:14
a lot for tuning in . I hope
43:16
you have enjoyed this week's episode . If
43:19
you like the show , please share , rate
43:21
and review . It helps more
43:24
listeners discover those stories
43:26
. You can find the links to all
43:28
the platforms the show appears on on
43:30
our website devjourneyinfo
43:33
slash subscribe
43:35
. Talk
43:38
to you soon .
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More